gis risk-based routes for hazardous shipments by roads
TRANSCRIPT
Journal of Traffic and Logistics Engineering Vol. 5, No. 1, June 2017
©2017 Journal of Traffic and Logistics Engineering 30doi: 10.18178/jtle.5.1.30-34
GIS Risk-Based Routes for Hazardous Shipments
by Roads
Rulla Saad Sharjah Research Academy, Government of Sharjah, UAE
Email: [email protected]
Khaled Hamad, Abdallah Shanableh, and Wameed Radeef Department of Civil and Environmental Engineering, University of Sharjah, Sharjah, UAE
Email: [email protected], [email protected], [email protected]
Abstract—Hazardous materials (HAZMATs) are products
that pose risk to health, property, and safety. Because of its
concentration, quantity, infectious or physical
characteristics. Such material can cause significant health
problems when released or spilled. Given all kinds of risks
to public health caused by hazardous material, transporting
these materials is a major concern for the public, authorities,
institutions, and professionals. The risk associated with
HAZMAT routing and its possibility of release makes it
different from the conventional vehicle routing method. The
main goal of this paper is to calculate the traditional risk to
find the least risky routes for transporting HAZMAT
between an origin-destination pair. This is accomplished
through two major steps: (1) identifying and computing
HAZMAT risk using traditional risk method on a segment-
by-segment basis, and (2) finding the least-risky HAZMAT
routes using shortest path algorithm in a GIS environment.
In this paper, the traditional risk is assumed to be the
product of the dose and the population exposed. Moreover,
the segment risk is defined as the product of the incident
probability and the population exposed, which can be used
to obtain the HAZMAT risk along a route. This risk model,
often termed as “traditional risk” measure, which has been
used in numerous HAZMAT routing studies that are
mentioned later in this paper.
Index Terms—Traditional risk, GIS, HAZMAT, origin-
destination, routes
I. INTRODUCTION
HAZMATs transportation by roads is of great concern
to any community because of the risks associated with
transporting these materials. Community concerns are
always related to accidents, noise, pollution quality of life,
and property value. In terms of transporting HAZMAT,
technical experts are not always trusted by the
community to provide complete information about the
level of associated risk. Although HAZMAT is
transported safely throughout any city daily, people
remain concerned about any potential release of
HAZMATs and the direct or indirect consequences to
health and environment.
Manuscript received February 24, 2017; revised June 13, 2017.
Many researchers proposed to use risk analysis to find
HAZMAT routes. Some of them calculated the minimum
population exposed to risk, the minimum societal risk and
minimum accident probability without using GIS [1]-[4].
Other researchers used GIS to calculate the minimum risk
[5]-[10]. As have been seen in literature, HAZMAT
routing is typically performed using traditional risk
method and shortest path algorithm. A widely-used
definition of risk is the product of the probability of an
incident and its associated consequence. The
consequences can range from fatalities to infrastructure
and environmental damages. Therefore, risk is typically
calculated based on exposed population that could be
endangered from HAZMAT accident or release. However,
HAZMAT routing is typically performed using
traditional risk calculation.
All models presented in the literature performed
routing based on traditional risk calculation. Although all
researchers used different travel impedances, yet still the
minimum risk was the utmost travel impedance that has
been applied. Furthermore, very few applied the
methodology on a segment-by-segment basis as [10].
Just to verify and differentiate between the study that
was done by [10] and this paper, they developed a study
to provide a comprehensive risk-based framework of
vehicle routing to account for various objectives
representing different perspectives. The objectives were:
travel distance, delay risk, travel cost, accident risk,
vehicle CO emissions risk, vehicle noise emissions risk
and greenhouse gas (Co2) risk. First, the mentioned
objectives were calculated using the comprehensive risk-
based approach which incorporates three main variables:
dose, response factor and exposed population. Equation
(1) shows the risk model used:
v (k) = (1)
where:
v(k) = comprehensive total risk
ri(k) = risk associated with objective i
wi = importance weight corresponding with ri(k)
Next, the analytical hierarchy process was used to find
the relative importance of each objective to the problem.
After that, the individual risk-based objectives are
Journal of Traffic and Logistics Engineering Vol. 5, No. 1, June 2017
©2017 Journal of Traffic and Logistics Engineering 31
combined in a single measure (i.e., combined risk). Then,
the individual objectives, in addition to the combined risk,
were defined as travel impedances to start the routing
analysis in GIS. The proposed methodology was applied
to City of Sharjah roadway network. The results were
optimal routes for a defined origin-destination pairs,
taking into consideration each risk-based objective
individually in addition to the risk as travel impedances.
This paper summarizes the work completed to find best
routes for transporting HAZMAT in the City of Sharjah
as a case study using risk-based analysis implemented in
a GIS environment.
II. MATERIALS AND METHOD
To build a model that can find the best HAZMAT
routes for a roadway network, the required data must be
first prepared and then input to the system. After that, the
risk is calculated for each roadway segment using the
traditional method for calculating risk. Then, the best
routes are found using a shortest path algorithm
implemented in a GIS environment. A graphical
representation of the proposed methodology to find the
best HAZMAT routes is shown in Fig. 1.
Prepare and Input Data
Module I – Identify and Compute
HAZMAT Risk on a Segment-by-
segment Basis
Module II - Find Least Risky Routes
Using GIS
Generate Maps for Best
Routes
Figure 1. Methodology framework
There are two types of input data that must be
collected. The first input data is the roadway network
data including: segment length (kilometers), traffic
volume (vehicle/hour), truck accident rate, and
probability of HAZMAT release. The second input data is
the population data which is used to calculate the number
of people exposed to HAZMAT risk. It is worth
mentioning that all data preparation were performed in a
GIS environment to produce the end results in an
implementable format. Preparing the roadway network
data in GIS proved to be challenging because conducting
routing analysis in GIS requires a special type of
topological linear network data, which allows for traffic
or any other commodity to flow on the network,
especially enabling vehicles to make turns at intersections
(nodes). If roadway segments, links or edges, are not
digitized properly, errors will appear when attempting to
use the shortest path algorithm in the GIS. Most of these
errors occur as a result of inaccurate digitizing of the data
in the original CAD file. Therefore to get accurate results,
it is very important to fix such errors if they exist in the
roadway network.
To demonstrate the applicability and accuracy of the
proposed model, it was used to select the best HAZMAT
route for the city of Sharjah roadway network, UAE.
Most of the roadway network data were brought from
Sharjah Road and Transport Authority, while the
population data were obtained from Sharjah Department
of Planning. However, some other required data were not
available such as: truck accident rates and probability of
HAZMAT release. So, a random variable function was
generated to estimate them. Based on the literature, the
truck accident rate was assumed to be in the range of
0.43x10-6 - 2.5x10-6 and the probability of release was
assumed to be in the range of 6.2x0-3 - 9.0x10-3. Fig. 2
demonstrates the traditional risk methodology that was
applied in this study.
Risk
Dose Exposed
Population
Number of
Accidents
Probability
of ReleaseX
X
Impacted
AreaPopulation
Density
Truck
Accident
Rate
Vehicle
Kilometers
Travel
X
X
Figure 2. Traditional risk components
The data in Fig. 3 and Fig. 4 show the roadway
network and the population blocks respectively for the
City of Sharjah, UAE.
Figure 3. Roadway network for the emirate of Sharjah
Figure 4. Population blocks for the emirate of Sharjah
Journal of Traffic and Logistics Engineering Vol. 5, No. 1, June 2017
©2017 Journal of Traffic and Logistics Engineering 32
The traditional risk in numerous HAZMAT routing
studies such as [11]-[9] was calculated by using equation
(1). Dose is defined as the quantity of risk causing agent,
which is the accident that happens because of the release
of HAZMAT. It can be calculated by using equation (3).
On the other hand, the probability of HAZMAT release is
defined as the probability of an incident involving a
vehicle carrying HAZMAT and the release of a
HAZMAT [9]. In the case study, it was assumed as a
random number that varies from 6.2*10-3 to 9.0*10-3 [8].
Risk = Dose x Exposed Population (2)
Dose = Number of Accidents * Probability of Release (3)
The number of accidents on a roadway segment was
estimated by using equation (4).
Number of Accidents = Truck Accident Rate * Vehicle
Kilometer of Travel (VKT) (4)
The exposed population (EP) is an estimate of the
population that is influenced by the consequences of
general movement, accidents and the release of
HAZMAT. The exposed population is assumed to be the
multiplication of the population density and the impacted
area as shown in equation (5).
EP = Population Density * Impacted Area (5)
In order to be able to calculate the exposed population,
the population density and the impacted area must be first
determined. In the case study, the impacted area was
chosen for one type of HAZMAT (which is flammable
liquid), and was assumed to be a flammable liquid with
an impact area = 0.5 mile [7]. The exposed population
was obtained by calculating the population density in
each population block as represented in Fig. 5. Then, a
road buffer of 0.5 miles was created along each road
segment using the buffer tool. After that, the resulted
buffers were intersected with the population blocks in
order to define the affected population, as seen in Figure
5. Each buffer segment can intersect with more than one
population block. The highlighted area is a part of each
population block, such that the line segment will take
attributes from the four population blocks; for that, the
average population density and the average impacted area
is determined.
Figure 5.
Representation of road segment buffer and the surrounding
population blocks
Thus, the result of the intersection is pieces of semi-
circles that must be summed up as one polygon for each
population block by using the dissolve tool in GIS. The
data in Fig. 6 show the results of the dissolve for the case
study. Finally, the population density and the area of road
buffer are determined and joined together to calculate the
exposed population.
Figure 6. Map showing the result of dissolve
III. RESULTS AND DISCUSSION
To demonstrate the applicability and accuracy of the
proposed traditional risk-based method for HAZMAT
routing, it was applied to find best routes (i.e., least risky)
for a real-world roadway network of the City of Sharjah
as a case study. While we can find best HAZMAT routes
between any two points (origin-destination) in the
network, we are only showing the results from two
selected routes in the city as examples.
Figure 7. Origin-destination 1: From University City to Al Khan
The first example transports HAZMAT from
University City to Al-Khan. Fig. 7 shows the three best
routes obtained based on three impedances that are:
minimum traditional risk, minimum travel time and
minimum travel distance. Table I shows three resulting
best routes for origin-destination example 1 from
University City to Al Khan using the following three
performance measures: route length, travel time, and
traditional risk. As can be seen, route 1 (which is the
minimum traditional risk route) has the least risk among
the three routes with a value of 35.1; route 2 has the least
travel time with 763 seconds; whereas route 3 as the least
Journal of Traffic and Logistics Engineering Vol. 5, No. 1, June 2017
©2017 Journal of Traffic and Logistics Engineering 33
travel distance with 15.5 Km. It is hard to compare the
three shown routes, since each route depends on a
different criterion. But, one can easily see in Fig. 7 that
the minimum risk route diverts away from the heavily
populated areas.
TABLE I. PERFORMANCE MEASURES FOR ORIGIN-DESTINATION 1
Routes Traditional
Risk
Travel Time
(sec)
Route Length
(Km)
1. Minimum
Traditional Risk 35.1 797 16.2
2. Minimum Travel Time
305 763 16.1
3. Minimum
Travel Distance 75.1 768 15.5
The second example shows best routes to transport
HAZMAT from Sharjah Airport to Khalid Port. Fig. 8
shows the three best routes obtained based on three
impedances that are: minimum traditional risk, minimum
travel time and minimum travel distance. Table II shows
three resulting best routes for origin-destination example
2 from Sharjah Airport to Khalid Port using the following
three performance measures: route length, travel time,
and traditional risk. As can be seen, route 4 (which is the
minimum traditional risk route) has the least risk among
the three routes with a value of 87.4; route 5 has the least
travel time with 769.6 seconds; whereas route 6 as the
least travel distance with 15 Km. It is hard to compare the
three shown routes, since each route depends on a
different criterion. But, one can easily see in Fig. 8 that
the minimum risk-based route diverts away from the
heavily populated areas.
Figure 8. Origin-destination 2: From Sharjah Airport to Khalid Port
TABLE II. PERFORMANCE MEASURES FOR ORIGIN-DESTINATION 2
Routes Traditional
Risk Travel Time
(sec) Route Length
(km)
4. Minimum
Traditional Risk 87.4 958.2 20.5
5. Minimum
Travel Time 240.2 769.6 16.7
6. Minimum
Travel Distance 199.8 722.8 15
IV. CONCLUSION
This paper summarizes a proposed approach to find the
best routes (i.e., least risky) for transporting HAZMAT
based on minimum risk. This was accomplished through
two modules: identify and calculate HAZMAT risk on a
segment-by-segment basis, and finding the least risky
routes using GIS.
REFERENCES
[1] E. Erkut and V. Verter, “Modeling of transport risk for hazardous materials,” Institute for Operations research and the Management
of Sciences, pp. 624-642, 1997.
[2] F. Diernhofer, B. Kohl, and R. Horhan, “Risk assessment of transportation for the dangerous goods in austrian road tunnels,” in
Proc. Fourth International Symposium on Tunnel Safety and Security, Germany, 2010.
[3] D. Barilla, G. Leonardi, and A. Puglisi, “Risk assessment for
hazardous materials transportation,” Applied Mathematical Sciences, pp. 2295–2309, 2009.
[4] P. Leonelli, S. Bonvicini, and G. Spadoni, “New detailed numerical procedures for calculating risk measures in hazardous
materials transportation,” Journal of Loss Prevention in the
Process Industries, pp. 507-515, 1999. [5] K. El-Basyouny, D. Cheng, and L. C. Sayed, “A risk/cost-based
algorithm for the routing of dangerous goods,” in Proc. Annual Conference of the Transportation Association of Canada,
Vancouver, British Columbia, 2009.
[6] V. Verter and B. Y. Kara, “A GIS-based framework for hazardous materials transport risk assessment,” Applied Mathematical
Sciences, pp. 2295–2309, 2001. [7] R. Ak and B. Bozkaya, “A proposed risk model and a gis
framework for hazardous materials transportation,” Istanbul
Technical University, Faculty of Management, Department of Industrial Engineering, Macka, Istanbul, 2008.
[8] Rahman and Lownes, “Risk assessment of hazardous materials transportation routes” Transportation Research Board, 2014.
[9] Y. Sun, “Risk-Based framework for freight movement analysis,”
Master thesis, Queensland University of Technology, Australia, 2002.
[10] Sahnoon, “Multi-Objective comprehensive risk-based vehicle routing analysis framework - Sharjah city as a case study,”
Master’s thesis, University of Sharjah, UAE, 2014.
[11] E. Erkut and V. Verter, “Modeling of transport risk for hazardous materials,” Operations Research, pp. 625–642, 1998.
Rulla R. Saad
is currently working as a Research Assistant at Sharjah
Research Academy –
Government of Sharjah. Engineer Rulla has more
than three years of work and academic experience in the field of traffic engineering, environmental engineering, sustainability, and
geographical information systems.
Khaled
I.
Hamad
is an Assistant Professor in the Department of Civil
& Environmental Engineering, University of Sharjah, United Arab Emirates. He received the B.Eng. from An-Najah National University,
Palestine, and M.Sc. and Ph.D. degrees all in civil engineering from the University of Delaware. His research interests are in Intelligent
Transportation Systems with emphasis on the applications of
Geographic Information Systems, artificial intelligence, and soft computing techniques
in transportation.
Abdallah Shanableh
is a Professor of Environmental Engineering and
Director of the Research Institute of Sciences and Engineering at the
University of Sharjah, United Arab Emirates. He received his BSc in Civil Engineering from the University of Jordan, Jordan, and his MSc
and PhD from the University of Texas at Austin, USA. He served in a number of professional, academic and administration positions in the
USA, Jordan, Australia, Japan and the USA.
Journal of Traffic and Logistics Engineering Vol. 5, No. 1, June 2017
©2017 Journal of Traffic and Logistics Engineering 34
Wameed M. Radeef is currently working as an Academic Coordinator at the Civil and Environmental Engineering Department of University
of Sharjah, UAE. Previously he was working as a Visiting Academic at
the same department. His research interests are in the area of waste management, biofuel production, sustainability and hazardous waste.